Psychiatric Disorders and Adverse Life Events as Predictors of Suicidality in a Clinical Sample of Low-Income Mexican Adolescents: Insights from a Machine Learning Approach

Background Suicide is a leading cause of adolescent death, with rates rising in low- and middle-income countries (LMICs) such as Mexico. Evidence on suicidality among vulnerable youth in psychiatric care in these settings remains scarce. Methods This cross-sectional study examined psychiatric disorders and adverse life events (ALEs) linked to suicidality in 127 low-income adolescents (ages 10–17) at Mexico’s national child psychiatric hospital. Participants completed validated psychiatric interviews and life event questionnaires. Four machine learning (ML) approaches—Classification Trees, Random Forests, XGBoost, and Logistic Regression—were used to identify key predictors. Results Across models, affective disorders were the strongest predictor of suicidality, followed by anxiety disorders, eating disorders, and ALEs involving family dysfunction, health problems, social adversity, and personal failures. Sexual abuse was reported by 25% of suicidal adolescents—twice the prevalence in non-suicidal peers. Peer conflict and weight-based stigma also distinguished groups, underscoring the role of social factors. Random Forests achieved the best overall predictive balance, though model performance was moderate. Findings reveal a complex interaction between psychiatric and psychosocial risks, and the value of ML in modeling these dynamics. Conclusions This is the first study to apply ML to suicidality in a clinical adolescent sample in Mexico. Results emphasize the need for culturally tailored, multidimensional prevention strategies addressing both psychiatric symptoms and the broader social determinants of mental health. Directly adopting models from high-income contexts without adaptation risks overlooking key contextual factors shaping suicidality in LMIC youth